Ribosomal Protein P0 Promotes Potato Virus A Infection and Functions in Viral Translation Together with VPg and eIF(iso)4E

Abstract

We report here that the acidic ribosomal protein P0 is a component of the membrane-associated Potato virus A (PVA) ribonucleoprotein complex. As a constituent of the ribosomal stalk, P0 functions in translation. Although the ribosomal stalk proteins P0, P1, P2, and P3 are all important for PVA infection, P0 appears to have a distinct role from those of the other stalk proteins in infection. Our results indicate that P0 also regulates viral RNA functions as an extraribosomal protein. We reported previously that PVA RNA can be targeted by VPg to a specific gene expression pathway that protects the viral RNA from degradation and facilitates its translation. Here, we show that P0 is essential for this activity of VPg, similar to eIF4E/eIF(iso)4E. We also demonstrate that VPg, P0, and eIF(iso)4E synergistically enhance viral translation. Interestingly, the positive effects of VPg and P0 on viral translation were negatively correlated with the cell-to-cell spread of infection, suggesting that these processes may compete for viral RNA.

INTRODUCTION

Viruses have acquired mechanisms by which they efficiently recruit the host translational machinery. Viral RNA translation is sometimes accompanied by the repression of host mRNA translation (1–4) by strategies that frequently involve the virus-induced modification of translation initiation factors. An alternative strategy is illustrated by a Cauliflower mosaic virus (CaMV) protein, which associates directly with the ribosome and regulates its function (5). Many host proteins that function in conventional translation also participate in viral RNA replication (6). For example, the replicase of bacteriophage Qβ contains four host proteins connected to translation, i.e., ribosome-associated HF-1, ribosomal S1, and elongation factors EF-Tu and EF-Ts (7). The eukaryotic homolog of EF-Tu is eEF1A, a protein frequently associated with viral replicases (8). Different subunits of eIF3 have also been shown to be functional components of both the Brome mosaic virus (BMV) and Tobacco mosaic virus (TMV) replicases (9–11).

Studies showing that only replicated viral RNAs are efficiently translated have indicated that positive-stranded RNA virus translation and replication are functionally coupled (12–14). The process of RNA replication is associated with virus-induced cellular membrane structures called viral replication complexes (RCs) (15). The model virus of this study, Potato virus A (PVA), is a positive-stranded RNA virus belonging to the genus Potyvirus. Several host proteins that function in translation have been associated with RCs of Turnip mosaic virus (TuMV) (genus Potyvirus), including eIF(iso)4E, PABP, and eEF1A (16–18). Since both PABP and eEF1A interact with the TuMV RNA-dependent RNA polymerase (RdRp), PABP and eEF1A may function in RNA replication (18, 19). Most recessive genes conferring resistance to infection by potyviruses are alleles of eIF4E and eIF(iso)4E, and successful infection commonly requires compatible interactions between viral VPg and eIF4E and eIF(iso)4E (20).

Ribosomes are essential for translation. In plants, they contain a structure referred to as a ribosomal stalk (21), which forms a lateral protrusion from the ribosome and is composed of the acidic proteins P0, P1, and P2 (22) and the plant-specific protein P3 (23, 24). P proteins are also present in non-ribosome-associated pools (25–27). P0 functions as a scaffold for the stalk structure by interacting with 28S rRNA, and the remaining P proteins assemble to form the extended stalk via interactions with P0. P0 is a protein conserved across kingdoms, as shown by the ability of P0 genes from worm, mammals, and protozoa to functionally complement Saccharomyces cerevisiae lacking endogenous P0 (28–30). P0 is the only essential P protein for both in vitro translational activity of yeast ribosomes and cell survival (31, 32). Ribosomal stalk proteins can affect several aspects of ribosome function, including translational capacity, polysome pattern, and ribosomal subunit joining (33, 34). P proteins are also regarded as having different effects on the translation of distinct mRNAs in yeast (31). In this study, we show that ribosomal P proteins are important for PVA infection of Nicotiana benthamiana. We found that P0, a viral RNP component, has functions distinct from those of the other P proteins in regulating PVA RNA expression.

MATERIALS AND METHODS

Plants.

Nicotiana benthamiana plants were grown in a greenhouse at 22°C for an 18-h day period and at 18°C for a 6-h night period and used for experiments at the 4- to 6-leaf stage.

Protein analysis.

Viral RNP complexes were purified from infected plants, and P0 was identified by proteomic tools as described previously (35). Ribosomes were isolated as described previously (36), except that the phosphatase inhibitors were omitted. P proteins were detected by Western blot analysis using a human autoimmune disease serum against ribosomal P antigen (catalog no. HP0-0100; Immunovision).

Viruses, plant overexpression, and gene silencing constructs.

PVA and firefly luciferase (FLUC) constructs were described previously (37). A P0 plant expression vector was constructed by generating a Gateway Cloning Technology (Invitrogen)-compatible cDNA of Arabidopsis thaliana 60S acidic ribosomal protein P0 (RPP0C) (GenBank accession no. NM_111960) by PCR. The cDNA was inserted into pMDC32 (38), pGWB17, and pGWB18 (39) via pDONR/Zeo (Invitrogen), using standard Gateway cloning. An eIF(iso)4E Gateway-compatible PCR product was recombined via pDONR/Zeo (Invitrogen) into pGWB18. The GUS and VPg plant expression constructs were described previously (40). P-protein-silencing vectors were constructed by generating Gateway-compatible N. benthamiana P-protein cDNA fragments, which were inserted into pHELLSGATE8 (pHG8) (41) via pDONR/Zeo; empty pHG8 was used as a control. The silencing constructs for eIF4E and eIF(iso)4E were described previously (40). All plant expression vectors were used to transform Agrobacterium tumefaciens strain C58C1/pGV2260.

Gene silencing by transient expression of hairpin RNA.

The method used for transient Agrobacterium-mediated silencing was described previously (42). Agrobacterium cells carrying hairpin vectors (pHG8) with gene-specific inserts were infiltrated into leaves. Hairpins were either cotransformed with wild-type (wt) PVA or pretransformed 4 days before inoculation of mutant PVAs. Gene silencing was verified by reverse transcription (RT)-PCR. Here, total RNA was extracted from plant leaves 4 days after infiltration of Agrobacterium cells carrying hairpin constructs. Total RNA was treated with DNase I, and cDNA was synthesized by using Moloney murine leukemia virus (M-MLV) reverse transcriptase and oligo(dT) primers. The same primers that were used to generate the sequences for cloning of cDNAs into pHG8 were used for PCR amplification.

Analysis of infection foci by fluorescence microscopy.

Agrobacterium cells carrying green fluorescent protein (GFP)-tagged PVA were infiltrated into plant leaves at an optical density at 600 nm (OD₆₀₀) of 0.0005. Exogenous proteins or RNA hairpins were coexpressed by coinfiltrating Agrobacterium cells carrying the respective expression cassettes. Infection foci were examined by using a Zeiss Axio Scope.A1 microscope and a 2.5× or 10× objective. The area of individual infection foci and the percentage of infected tissue were determined by using microscope software. For each condition, data on infection foci were collected from a minimum of three individual plants and 50 foci.

RLUC- and qPCR-based PVA infection assay.

Luciferase quantitation of virus-derived Renilla luciferase (RLUC) and control FLUC and quantitative RT-PCR (qPCR) were described previously(37, 40). Details of Agrobacterium OD₆₀₀s used to transform PVA and other expression cassettes, time points of data collection, and other relevant information are given in the respective figure legends and the text. Error bars represent the standard deviations.

Nucleotide sequence accession numbers.

Data have been deposited in the GenBank database under accession no. NM111960 (A. thaliana RPP0C); FN666434 [Nicotiana tabacum eIF(iso4)E]; and JN227614 (P0), JN227615 (P1), JN227616 (P2), and JN227617 (P3) (N. benthamiana silencing fragments).

RESULTS

Ribosomal P0 copurifies with viral replication proteins from infected plants.

To identify host proteins participating in PVA infection, two viral replication proteins, VPg and RdRp, were expressed as Strep III-tagged fusion proteins from their native positions in the viral genome and used to affinity purify membrane-associated viral protein complexes from infected Nicotiana benthamiana plants. The composition of these complexes was analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS), which showed that HSP70, viral RdRp, NIa, and VPg were all found in both the VPg- and RdRp-tagged samples (35). In this study, we report the identification of a protein homologous to Arabidopsis thaliana 60S acidic ribosomal protein P0 (P0) (GenBank accession no. AAL07229). Western blot analysis using human anti-P autoimmune disease serum showed that P0 was specifically copurified with both Strep III-tagged VPg and RdRp but was not visible in the control sample derived from leaves infected with nontagged PVA (Fig. 1A).

Fig 1.

Ribosomal P0 is a component of membrane-associated viral RNPs. (A) Western blotting showing the presence of P proteins in the RNP complexes obtained by affinity purification, using Strep III-tagged PVA RdRp (RdRp^SIII) and Strep III-tagged VPg (VPg^SIII), from membrane fractions of infected N. benthamiana leaves. A sample purified from plants infected with wt PVA was used as a control. (B) Western blotting of P proteins associated with isolated ribosomes from N. benthamiana leaves. Human anti-P autoimmune disease serum was used to detect P proteins in panels A and B.

The serum used to detect P0 also recognizes the ribosomal stalk proteins P1, P2, and P3 via their homologous C termini (23, 43). These proteins are small (10 to 15 kDa) and may correspond to the weakly detected lower-molecular-weight proteins (Fig. 1A). The patterns of viral RNP- and ribosome-associated P proteins differed. A Western blot analysis with the human anti-P autoimmune disease serum of a ribosome-enriched fraction derived from noninfected N. benthamiana leaves yielded several proteins (Fig. 1B) with a pattern corresponding to that of maize ribosomal P proteins (44). In contrast, the viral RNP complex contained more P0 than the other P proteins. None of the other P proteins was detected by LC-MS/MS analysis, providing further evidence that they were not as abundant as P0 in the purified viral RNP.

Exogenous P0 promotes PVA infection.

We inserted the Renilla luciferase (RLUC) reporter gene into PVA infectious cDNA to enable sensitive quantification of viral infection (Fig. 2) (37). To analyze the role of P0 in PVA infection, we constructed silencing and overexpression constructs for P0. P0 was transiently silenced by expressing a P0-targeting RNA hairpin in plant leaves using Agrobacterium infiltration, with the control being the empty hairpin vector. Western blot analysis using anti-P serum showed that silencing reduced the expression level of P0 (Fig. 3A). In contrast, when P0 was overexpressed, no obvious P0 accumulation could be detected (Fig. 3B), similar to findings in yeast cells, which do not accumulate P0 protein despite increased levels of P0 mRNA (45). We therefore verified ectopic P0 expression using N- and C-terminally fused myc tags and found that these proteins could be detected by using anti-myc IgGs (Fig. 3C). To address the role of P0 in infection, exogenous P0, with and without myc tags, was expressed along with RLUC-tagged wt PVA, and the effect on RLUC activity at 7 days after infection (DAI) was analyzed (Fig. 3D). GUS was expressed as a control for P0. Exogenous expression of both tagged and nontagged P0 increased viral RLUC activity by 3-fold, whereas silencing of P0 reduced viral RLUC activity by 3-fold (Fig. 3D).

Fig 2.

Schematic representation of the constructs used in this study. The transcription CaMV 35S promoter and the nopaline synthase (NOS) and octopine synthase (OCS) terminators are in gray, and cDNAs are in white. Viral constructs lacking either the PVA 5′- or 3′-UTR were derived from the PVA^ΔGDD construct. Both monocistronic RLUC and monocistronic GUS were expressed from similar constructs to FLUC and VPg. eIF(iso)4E was expressed from a similar construct as P0 in the pGWB18 vector backbone.

Fig 3.

Ribosomal P0 promotes PVA infection. (A) P0 was silenced by expressing an RNA hairpin (hp) that targets P0 transcripts (hpP0), followed by Western blotting using human anti-P autoimmune disease serum directed against the ribosomal P antigen. The empty hairpin vector was expressed as a control (hp −). (B) Exogenous P0 was expressed, and P0 levels were analyzed by Western blotting as described above for panel A. GUS was expressed as a control. (C) Exogenous expression of myc-tagged P0 followed by Western blotting. P0^N and P0^C have N- and C-terminal myc fusions, respectively. The membrane stained for total protein by Ponceau S is shown at the position of the RUBISCO large subunit as a loading control in panels A to C. (D) Exogenous P0 with or without myc tags or GUS (control) was coexpressed with wt PVA. In parallel, P0 was silenced, and the level of luciferase activity derived from RLUC-tagged wt PVA was determined 7 days after infection (DAI). The Agrobacterium OD₆₀₀ values used for infiltration were as follows: 0.0005 for wt PVA and 0.5 for P0, GUS, hp −, and hpP0. (E) Fluorescence microscopy of the average area of infection foci formed by GFP-tagged wt PVA at 3 DAI. (F) Representative infection foci from leaves expressing GUS and P0. Scale bar, 200 μm. The number of analyzed foci was >100 for both GUS and P0. *, P < 0.05; **, P < 0.01 (determined by Student's t tests).

We also estimated the average area of individual infection foci formed by wt GFP-tagged PVA in the presence of exogenous P0 or GUS at 3 DAI (Fig. 3E). GFP fluorescence was detected in numerous cells within separate foci at this time, showing that infection had already spread by cell-to-cell movement (Fig. 3F). Expression of exogenous P0 increased the average area of infection foci compared with the control. It is important to emphasize that very low concentrations of Agrobacterium (OD₆₀₀ = 0.0005) were used to transform RLUC/GFP-tagged wt PVA (Fig. 3D to F), resulting in the initiation of 1 infection focus per mm² (e.g., see Fig. 4B). This low density reduced any possible effects due to the Agrobacterium-based inoculation method and left space for the infection to spread by cell-to-cell movement. However, it was necessary to measure the diameter of the infection foci at 3 DAI, as at 4 DAI, many foci will have coalesced (data not shown). Separate infection foci in which cell-to-cell movement had occurred by 3 DAI were also achieved with TuMV under similar inoculation conditions (46).

Fig 4.

Ribosomal P proteins are important for PVA infection. (A) RLUC activity of wt PVA was determined at 3, 6, and 9 DAI during silencing of P proteins by using RNA hairpins (hp) targeting each P protein (hpP0, hpP1, hpP2, and hpP3). An empty hairpin vector was expressed as a control (hp −). The Agrobacterium OD₆₀₀ values used for infiltration are described in the legend of Fig. 3D. (B) Average numbers of infection foci per mm² 3 days later. Twelve areas of 4 mm² were analyzed under each condition by using fluorescence microscopy. (C) The average size of infection foci was determined. Over 100 foci were analyzed under each condition. *, P < 0.05; **, P < 0.01 (determined by Student's t tests). (D) Representative leaf tissue images of PVA-derived GFP fluorescence in infection foci during P-protein silencing at 3 and 6 DAI. Scale bar, 1 mm. (E) RT-PCR was used to verify silencing of P0 to P3 transcripts achieved by expression of RNA hairpins. The expected PCR products are indicated by arrows, and the size marker (M) is on the left.

Ribosomal stalk proteins are important for PVA infection.

The results described above showed that P0 is important for PVA infection. Since P0 is essential for cellular translation as part of the ribosomal stalk structure, altered P0 homeostasis may affect PVA infection through effects on either the ribosomal stalk or other P0 functions. To analyze the possible role of the ribosomal stalk in PVA infection, RNA hairpins were expressed to silence the other stalk P proteins (P1, P2, and P3) as well as P0, and wt PVA RLUC activities were analyzed at 3, 6, and 9 DAI (Fig. 4A). Although silencing of any P protein reduced viral RLUC activity, the latter was higher during P0 silencing than during P1, P2, or P3 silencing at 9 DAI.

The infection foci of GFP-tagged PVA were also analyzed by fluorescence microscopy. First, the number of GFP foci was calculated at 3 DAI (Fig. 4B), a time point when most foci were isolated clusters of cells showing PVA-derived GFP fluorescence (e.g., see Fig. 3F). Silencing of P proteins did not alter the amount of initially infected cells compared with the control, indicating that the reduction in viral RLUC activity during P-protein silencing was not due to reduced initiation but to reduced progression of infection. Moreover, silencing of the P proteins resulted in a slight reduction in the average areas of infection foci at 3 DAI (Fig. 4C). At 6 DAI, however, the leaves of both the control and P-protein-silenced plants were fully infected, as GFP fluorescence was uniform throughout the leaves (Fig. 4D). Fluorescence intensity, as well as viral RLUC activity, was strongest in the control (Fig. 4A). Together, these results show that the silencing of ribosomal stalk proteins affects PVA infection by reducing the cellular level of infection but has little effect on the cell-to-cell spread of infection. RT-PCR verified silencing of the target transcripts (Fig. 4E).

P0 has functions that are distinct from those of other P proteins in PVA infection.

Coordination of the potyvirus genome between translation, replication, and other cellular functions is complex, as these stages are coupled, and inhibition of virus infection due to the silencing of P proteins may result from disruption of any of those processes. We therefore utilized different RLUC-tagged PVA mutants to determine which viral process is affected under each experimental condition. PVA^CPmut is a virus with a coat protein (CP) mutation that undergoes replication in inoculated cells but is incapable of virus particle assembly and cell-to-cell and systemic movement (37). PVA^ΔGDD carries a mutation in the RdRp that inhibits replication (37); it therefore does not carry VPg at its 5′ end, differentiating it from authentic viral RNAs. The validity of the use of this construct for studying PVA translation is addressed in Discussion.

P proteins were silenced by transforming RNA hairpins 4 days prior to inoculation of mutant PVAs, along with firefly luciferase (FLUC) as an internal control for Agrobacterium-mediated nonviral expression. PVA RLUC activity and RNA accumulation were assayed at 3 DAI (Fig. 5A and B). Importantly, the RNA level was 10-fold higher for PVA^CPmut than for PVA^ΔGDD in the control plants, indicating that most of the PVA^CPmut RNA was derived from replication. PVA^CPmut RNA levels were markedly reduced by silencing of P1, P2, and P3, suggesting reduced replication. This may have been due to reduced translation, as both PVA^ΔGDD and FLUC RNA levels were unaffected, although the corresponding luciferase activities decreased (Fig. 5B and C). In contrast to the other P proteins, elevated PVA^CPmut RNA levels were detected after P0 silencing, suggesting that P0 can affect PVA RNA accumulation differently from the other stalk proteins and thus explaining why P0 was identified from membrane-associated viral RNPs (Fig. 1). The ratio of RNA to RLUC activity reflects the translation of accumulated RNA. The 4-fold increase in the ratio of PVA^CPmut RNA to RLUC during P0 silencing (Fig. 5D) suggests that despite the accumulation of PVA^CPmut RNA, it was not efficiently translated.

Fig 5.

P0 has functions distinct from those of other P proteins in PVA infection. (A to C) RLUC activity and RNA accumulation were determined for PVA^CPmut (A), PVA^ΔGDD (B), and FLUC (C) during P-protein silencing at 3 DAI. Hairpins were applied 4 days before PVAs. The Agrobacterium OD₆₀₀ values used for infiltration were as follows: 0.05 for PVA^CPmut and PVA^ΔGDD, 0.005 for FLUC, and 0.4 for hairpins. Luciferase activities and RNA amounts were analyzed by using the same samples. (D) Ratio of RNA amount to viral RLUC activity for PVA^CPmut, calculated by dividing the average relative values calculated from data presented in panel A. *, P < 0.05; **, P < 0.01 (determined by Student's t tests).

P0 can increase viral translation.

Since the expression of exogenous P0 increased wt PVA infection (Fig. 3D), we analyzed whether exogenous P0 expression affected RLUC-tagged PVA^ΔGDD and FLUC activities and RNA accumulation. Exogenous P0 increased RLUC activity but did not alter PVA^ΔGDD RNA levels (Fig. 6A), suggesting that P0 enhanced viral translation. FLUC activity and RNA levels did not respond to exogenous P0. The positive effect of exogenous P0 expression on RLUC accumulation was similar when derived from PVA^ΔGDD and wt PVA infection (Fig. 3D), indicating that P0 could promote wt infection through increased viral translation.

Fig 6.

P0 can increase viral translation. (A) Exogenous P0 was coexpressed with PVA^ΔGDD and FLUC, and luciferase activity and RNA amounts were assessed at 3 DAI. GUS was coexpressed as a control. The Agrobacterium OD₆₀₀ values used for infiltration were as follows: 0.05 for PVA^ΔGDD, 0.005 for FLUC, and 0.5 for GUS and P0. **, P < 0.01 (determined by Student's t tests).

The 5′-UTR is an important part of the viral RNA response to P0.

The viral untranslated regions (UTRs) appear to promote PVA RNA translation, as the removal of either UTR reduced viral protein accumulation (40). We analyzed the importance of viral UTRs in the P0 promotion of viral translation by expressing exogenous P0 with PVA^ΔGDD lacking either the 5′-UTR (PVA^{ΔGDDΔ5′UTR}) or the 3′-UTR (PVA^{ΔGDDΔ3′UTR}) and determined viral RLUC activity (Fig. 7A). RLUC activity hardly increased for PVA^{ΔGDDΔ5′UTR}, suggesting that the 5′-UTR is important for exogenous P0 to enhance viral translation. Exogenous P0 increased RLUC activity derived from PVA^{ΔGDDΔ3′UTR} more than that from PVA^ΔGDD. Hence, the 5′- and 3′-UTRs exerted opposite effects on viral RNA translation in the presence of exogenous P0. We further analyzed if monocistronic RLUC transcripts carrying the viral 5′-UTR could respond to exogenous P0 expression (Fig. 7B). Its RLUC activity was unaltered by exogenous P0 expression, suggesting that the viral RNA has additional properties required for P0 to increase its translation.

Fig 7.

The 5′-UTR is important for PVA RNA translation and the P0 response. (A) Exogenous P0 was expressed with PVA^ΔGDD or with PVA^ΔGDD lacking either its 5′-UTR (PVA^{ΔGDDΔ5′UTR}) or its 3′-UTR (PVA^{ΔGDDΔ3′UTR}), and RLUC activities were determined at 3 DAI. GUS was expressed as a control. Results are expressed as relative luciferase activity. The Agrobacterium OD₆₀₀ values used for infiltration were as follows: 0.05 for PVAs and 0.5 for GUS and P0. (B) Exogenous P0 or GUS was expressed with a monocistronic RLUC transcript carrying the PVA 5′-UTR (5′UTR^PVA), and RLUC activities were determined. The Agrobacterium OD₆₀₀ values used for infiltration were as follows: 0.05 for the PVA 5′-UTR and 0.5 for GUS and P0. (C) Actual RLUC activities and RNA levels for PVA^ΔGDD, PVA^{ΔGDDΔ5′UTR}, and PVA^{ΔGDDΔ3′UTR}. (D) Monocistronic RLUC constructs with a reference leader (5′UTR^Ref) or the PVA 5′-UTR were transformed using an OD₆₀₀ of 0.01, and the RLUC activity and RNA amounts were quantified. *, P < 0.05; **, P < 0.01 (determined by Student's t tests).

Next, we correlated RLUC activities with PVA^ΔGDD, PVA^{ΔGDDΔ5′UTR}, and PVA^{ΔGDDΔ3′UTR} RNA levels (Fig. 7C). Deletion of the 5′-UTR reduced RLUC activity more than RNA levels compared with PVA^ΔGDD, whereas an absence of the 3′-UTR reduced RLUC activity but not RNA levels, showing that both UTRs are important, especially for viral RNA translation. We also compared RLUC activities and RNA levels of monocistronic RLUC transcripts fused with two different 5′-UTRs, a cloning vector-derived random leader sequence (5′-UTR^Ref) and the PVA 5′-UTR (Fig. 7D). The viral 5′-UTR did not enhance either the accumulation or translation of the RLUC transcript. These findings indicate that the viral 5′-UTR is essential for efficient viral translation and is required for a translational response of PVA^ΔGDD to exogenous P0.

P0 is required for VPg-enhanced viral translation.

We reported previously that VPg requires the PVA 5′-UTR to increase viral translation when expressed in trans but that the PVA 5′-UTR alone is not sufficient to transfer this response to a monocistronic transcript (40), similarly to our results for P0. We therefore analyzed whether P0 and VPg were connected in promoting viral translation. As the expressions of exogenous VPg (40) and P0 increase RLUC activity from both wt PVA and PVA^ΔGDD, we used PVA^ΔGDD in this expression analysis to exclude any effects of virus replication and movement on viral protein accumulation. Either VPg or GUS (control) was coexpressed with PVA^ΔGDD during P-protein silencing (Fig. 8). VPg increased viral RLUC activity 3-fold in control and P1-, P2-, and P3-silenced leaves but not during P0 silencing, showing that of these P proteins, only P0 was required to increase viral translation.

Fig 8.

VPg requires P0 to enhance viral translation. VPg was coexpressed with PVA^ΔGDD during P-protein silencing, and the RLUC activities were quantitated at 3 DAI. Results are expressed as relative values. The Agrobacterium OD₆₀₀ values used for infiltration were as follows: 0.05 for PVA^ΔGDD and 0.5 for GUS and VPg. *, P < 0.05; **, P < 0.01 (determined by Student's t tests).

PVA translation is increased further by P0 and VPg together.

To further analyze the connection between P0 and VPg in viral translation, PVA^ΔGDD was coexpressed with P0 and/or VPg, and RLUC activity and RNA levels were determined at 3 DAI (Fig. 9A). As before, P0 increased RLUC activity but not the PVA^ΔGDD RNA level (Fig. 6A), whereas VPg increased both (40). Further increases in RLUC activity and viral RNA levels were observed when P0 and VPg were coexpressed. The levels of monocistronic FLUC RNA and FLUC activity remained essentially unaltered (Fig. 9B), showing that the response was specific for PVA^ΔGDD RNA.

Fig 9.

P0 and VPg act synergistically to increase PVA translation. (A and B) VPg and/or P0 was coexpressed with PVA^ΔGDD (A) and control FLUC (B), and luciferase activity and RNA levels were quantified at 3 DAI. GUS was expressed as a control. The Agrobacterium OD₆₀₀ values used for infiltration were as follows: 0.05 for PVA^ΔGDD, 0.005 for FLUC, and 0.25 for P0 and VPg. Agrobacterium carrying the uidA gene (GUS) was used to adjust all infiltrated samples to the same final OD₆₀₀. (C) GUS/P0/VPg coexpressions, as in panels A and B, in leaves systemically infected with RLUC-tagged wt PVA. The lower leaves were inoculated with PVA, and 7 days later, the systematically infected upper leaves were infiltrated with Agrobacterium carrying GUS/P0/VPg expression cassettes. Viral RLUC activity was determined 3 days later, corresponding to 10 days after PVA inoculation. *, P < 0.05; **, P < 0.01 (determined by Student's t tests).

To determine whether P0 and VPg together could affect natural infection, lower leaves of plants were inoculated with wt PVA, and after 7 days, P0 and VPg were expressed by Agrobacterium-mediated transformation in systemically infected upper leaves. Viral RLUC activity was determined 3 days after infiltrating P0 and VPg expression constructs into systemically infected leaves, equal to 10 days after inoculation of the plants with PVA (Fig. 9C). Expression of either exogenous VPg or P0 increased viral RLUC activity. Exogenous P0 increased systemic infection by wt PVA (Fig. 3D) and PVA^ΔGDD (Fig. 6A and 7A), further indicating that the P0 response functions in bona fide PVA infection. Expression of exogenous VPg increased viral RLUC activity in systemically infected leaves but less than that observed for PVA^ΔGDD (Fig. 9A). The synergistic effect of VPg and P0 on PVA gene expression could not be detected here.

Exogenous VPg increases viral translation together with eIF(iso)4E and P0 but reduces the spread of infection.

We showed previously that exogenously expressed VPg is unable to enhance viral translation when both eIF4E and eIF(iso)4E are silenced (40). To extend this finding, we cosilenced eIF4E and eIF(iso)4E and analyzed RLUC activities and RNA levels of RLUC-tagged wt PVA at 9 DAI (Fig. 10A). Both RLUC activity and viral RNA levels were reduced during eIF4E/eIF(iso)4E silencing and during P0 silencing (Fig. 10A). As both P0 and eIF4E/eIF(iso)4E are required for normal PVA infection and for a VPg-mediated boost in translation, the effects of different combinations of eIF(iso)4E, VPg, and P0 on wt PVA RLUC activity were determined at 3 DAI (Fig. 10B). To our surprise, RLUC activity was not increased by VPg or eIF(iso)4E and was increased only slightly by the combination of VPg, P0, and eIF(iso)4E but was markedly increased by VPg plus eIF(iso)4E. We therefore analyzed the effects of VPg, P0, and eIF(iso)4E expression on the development of infection foci by GFP-tagged wt PVA by determining the average area of infection foci at 3 DAI (Fig. 10C), at which time cell-to-cell movement had already occurred (see also Fig. 3F and 4D). Exogenous VPg expression reduced the intercellular spread of infection (Fig. 10C), an effect even more pronounced when the proportion of infected to noninfected tissue was determined at 4 DAI (Fig. 10D and F). Interestingly, coexpression of P0 with VPg reduced cell-to-cell infection spread even more than VPg alone (Fig. 10D), despite an increase when P0 was expressed alone (Fig. 3F). Also, the positive effect of VPg and eIF(iso)4E coexpression on PVA RLUC activity was overpowered when P0 was coexpressed. Importantly, the total number of infection foci remained similar under these conditions, showing that altering the PVA infection initiation rate was not responsible for the observed effects (Fig. 10E).

Fig 10.

Exogenous VPg increases viral translation with eIF(iso)4E and P0 but reduces spread of infection. (A) P0 and both eIF4E and eIF(iso)4E were silenced by expression of the corresponding RNA hairpins (hpP0 and hp4Es), and the levels of luciferase activity and RNA derived from RLUC-tagged wt PVA were determined at 9 DAI. The empty hairpin vector was expressed as a control (hp −). The Agrobacterium OD₆₀₀ values used for infiltration were as follows: 0.0005 for wt PVA, 0.4 for hp − and hpP0, and 0.2 each for hp4Es [hpeIF4E and hpeIF(iso)4E]. (B to E) eIF(iso)4E was coexpressed with RLUC-tagged wt PVA in different combinations with VPg and/or P0. RLUC activity (B) and the area of infection foci (C) were determined at 3 DAI, and the percentage of GFP-tagged wt PVA-infected tissue at 4 DAI (D) and the number of established GFP-tagged PVA infection foci at 3 DAI (E) were determined. (F) Representative pictures of GFP-tagged wt PVA-infected tissue at 4 DAI during GUS and VPg coexpression. Scale bar, 1 mm. (G and H) Effects of the same coexpressions on PVA^ΔGDD (G) and FLUC (H) luciferase activities. eIF(iso)4E, P0, and VPg each used an Agrobacterium OD₆₀₀ of 0.25 for infiltration, and GUS was used to adjust all infiltrated samples to the same final bacterial density. The other Agrobacterium OD₆₀₀ values used for infiltration were as follows: 0.0005 for RLUC/GFP-tagged wt PVA, 0.05 for PVA^ΔGDD, and 0.005 for FLUC. *, P < 0.05; **, P < 0.01 (determined by Student's t tests).

When wt PVA is infiltrated at a high OD₆₀₀, VPg increases viral RLUC activity (40). At an OD₆₀₀ of 0.05, PVA RNA expression is often initiated from adjacent cells, leaving less space for cell-to-cell spread of infection (37). At a 100-fold-lower OD, the size of infection foci was severely reduced by exogenous VPg expression (Fig. 10D), with the lack of increased wt PVA RLUC activities (Fig. 10B) likely being due to impaired cell-to-cell spread of infection. Exogenous VPg increased viral RLUC activity when expressed in systemically infected leaves (Fig. 9C), showing that this mechanism operates during authentic infection. As the VPg-mediated increase of RLUC activity was less than that for PVA^ΔGDD, intercellular spread was likely hindered. The synergistic effect of VPg and P0 coexpression was not detected, which may also be explained by restricted cell-to-cell movement. Finally, we analyzed how the same combinatory expressions affected PVA^ΔGDD RLUC activity (Fig. 10G). Viral RLUC activity was higher when VPg was coexpressed with eIF(iso)4E than when either was expressed alone, indicating that eIF(iso)4E and VPg increased PVA^ΔGDD RLUC activity synergistically. The highest RLUC activity was achieved by coexpressing VPg with both eIF(iso)4E and P0. Control FLUC activity was unaffected under these conditions (Fig. 10H), indicating that viral RNA has specific properties by which it responds to ectopic expression of VPg, P0, and eIF(iso)4E.

DISCUSSION

Virus infection relies on mechanisms that protect viral RNA from being degraded and guarantee efficient production of viral proteins in a cellular environment. We have shown here that ribosomal P0 is important for PVA infection by promoting viral translation along with eIF(iso)4E and the viral protein VPg. P0 functions as part of the ribosomal stalk, with all the acidic P proteins that constitute the ribosomal stalk being important for PVA infection. Our findings also suggest that P0 functions as an extraribosomal protein in PVA infection and regulates viral RNA functions.

Ribosomal stalk proteins are important for PVA infection.

Few previous reports have linked ribosomal stalk proteins to virus infection. Lassa virus Z protein interacts with the nuclear fraction of P0 (47). Assembly of the 60S ribosomal subunit occurs in the nucleus, and upon export to the cytoplasm, the subunit matures through processes including association of the stalk (48). This suggests that Z protein interactions probably did not involve ribosomes. L-A virus has been found to propagate to higher levels in yeast strains lacking P1 and P2, with the viral Gag protein being associated with P0 (49). We found that silencing of all P proteins reduced PVA infection substantially. Because translation and replication correlate positively for some animal positive-stranded RNA viruses (50, 51), the reduction in PVA infection following P-protein silencing may be due to reduced translation, resulting in reduced replication. Genome-wide RNA interference (RNAi) screening identified 72% of all ribosomal proteins whose knockdown attenuated infection by Drosophila C virus (DCV), including P0 and P2, indicating the importance of ribosomal proteins for DCV infection (52). The sensitivity of DCV infection to knockdown of ribosomal proteins was shown to be connected to a cap-independent internal ribosomal entry site (IRES) translation initiation mechanism. Similar to DCV, PVA is an RNA virus that potentially employs IRES-dependent translation mechanisms, as do other potyviruses such as Tobacco etch virus (TEV) and TuMV (53–55). TuMV and TMV (genus Tobamovirus) infections were highly reduced by knockdown of several ribosomal proteins (56). The knockdown of RPS2, RPS6, RPL7, RPL13, and RPL19 reduced the number of TuMV infection foci, regardless of whether the plants were inoculated with virus particles or Agrobacterium. In contrast, P-protein silencing did not reduce the number of infection foci and affected the spread of PVA only slightly. The other ribosomal proteins may therefore have roles distinct from those of the stalk proteins in potyvirus infection. Clearly, ribosomal protein homeostasis is important for infection by potyvirus as well as other viruses.

Extraribosomal regulation of PVA RNA by P0.

Our findings suggest that compared with other P proteins, P0 has additional functions. For example, P0 was a component of viral RNP complexes purified via both VPg and RdRp, whereas the other P proteins were not, suggesting that the latter were not prominent components of purified viral RNPs (Fig. 1). These viral RNP complexes may have been derived from RCs, as they were purified from membrane fractions containing potyviral RNA synthesis activity (57, 58). HSP70 is associated with potyviral RCs and the RdRp (19) and was present in our purified samples, together with viral RNA, RdRp, NIa, and VPg (35). Therefore, P0 may associate with viral RNP complexes at the sites of viral RCs, possibly as an extraribosomal protein.

The abundant accumulation of PVA^CPmut RNA during P0 silencing (Fig. 5A) suggests a functional role for P0 in PVA RNA regulation. This accumulation occurred only during silencing of P0 and not during silencing of the other P proteins, suggesting that, independent of stalk function, P0 can affect PVA RNA accumulation. Although replicating PVA^CPmut RNA accumulated in single cells in the absence of P0, this RNA was not translated efficiently (Fig. 5). P0 was required for wt PVA to reach normal levels of infection (Fig. 3D), most likely due to the function of P0 in viral translation. In particular, we found that P0 was the only P protein required for a specific viral translation mechanism involving viral VPg and eIF(iso)4E (Fig. 8 to 10). Ribosomal proteins have been shown to have extraribosomal functions, e.g., plant L10 in geminivirus infection (59) and S1 associated with bacteriophage Qβ replicase (7). Extraribosomal P0 is part of the mRNA-regulating human IMP1 and STAU1 mRNP complexes (60, 61) and may function in nucleus-associated DNA repair (62). Taken together with our findings, P0 likely regulates viral RNA via extraribosomal mechanisms.

P0 and eIF(iso)4E are essential for VPg to promote PVA RNA translation.

Several potential translational components are associated with RCs of TuMV, including eIF(iso)4E, PABP, eEF1A, and AtRH8 (16–18, 63), but their roles have remained elusive. They may function to regulate viral translation (64), as translation is coupled with replication (35) and RCs (65) during potyvirus infection. P0 is found in membrane-associated viral RNP complexes (Fig. 1) and promotes PVA translation, suggesting that it may be an additional potyviral RC-associated host protein involved in translation. Exogenous P0 expression increased both wt PVA infection and PVA^ΔGDD translation (Fig. 3, 7, 9, and 10). We reported previously that exogenous expression of VPg increases both the stability and translation of wt PVA and PVA^ΔGDD RNA and that simultaneous silencing of eIF4E and eIF(iso)4E did not decrease basal PVA^ΔGDD translation but was essential for increased VPg translation (40). Similarly, we showed that P0 is required for VPg-mediated enhancement of PVA RNA translation (Fig. 8) and that PVA^ΔGDD translation (Fig. 5B) was less sensitive to reduced P0 amounts than wt PVA infection (Fig. 3D). Higher RNA and VPg levels of replicating viruses may require more P0 and eIF4E/eIF(iso)4E, making their translation more sensitive to silencing of these proteins. P0 transcription is induced early in PVA infection (66), and TuMV infection upregulates eIF4E (67), suggesting that the demand for these proteins is increased during natural potyvirus infection. This was supported by results showing that the silencing of both of these host factors reduced PVA infection similarly (Fig. 10A).

During infection, VPg-linked viral RNA is produced within membrane-associated RCs. Our results using PVA^ΔGDD show that viral RNA does not have to be produced via replication in authentic RCs for VPg, P0, and eIF(iso)4E to efficiently promote its translation. The viral 5′-UTR is an important RNA element by which P0 and VPg increase PVA translation (Fig. 7A) (40). Potyvirus TEV has an IRES element in the 5′-UTR that can recruit ribosomes independently of eIF4Es via direct interactions with eIF4G (54). TEV VPg has been shown to be associated with the eIF4F protein complex consisting of eIF4E and eIF4G, reducing the affinity of eIF4F for the conventional cap structure found in the 5′-UTR of cellular mRNAs and facilitating the translation of transcripts carrying the TEV 5′-UTR (68). The presence or absence of a 5′ cap did not alter the ability of VPg to facilitate the translation of TEV 5′-UTR transcripts. By analogy, the capped 5′-UTR of PVA^ΔGDD RNA may qualify as a substrate for IRES translation. Furthermore, as wt PVA RNA responds to exogenous VPg by increased translation in systematically infected leaves (Fig. 9C), it can be concluded that 5′ VPg-linked PVA RNA also qualifies for VPg-mediated enhancement of PVA RNA translation, suggesting that this mechanism can operate during natural infection. Therefore, the mechanism of the VPg response may involve PVA 5′-UTR-driven IRES translation. Its significance may be reflected by the reduced infection observed during eIF4E/eIF(iso)4E and P0 silencing. eIF4E and eIF(iso)4E are critical host components by which potyviruses establish infection, by interacting with VPg (20). Although the mechanisms by which eIF4E and eIF(iso)4E act during potyvirus infection have been hard to determine, our results show that they are essential for VPg to increase PVA translation (40).

Coordination of viral RNA functions.

Although VPg enhanced wt PVA gene expression (Fig. 9C) (40), it restricted the spread of infection (Fig. 10D). Reduced viral spread likely masked the increase in viral gene expression levels during active cell-to-cell movement. A comparison of areas of infection foci (Fig. 10D) with RLUC activity (Fig. 10B) indicated that VPg also enhanced translation under these conditions. Our experiments assessing the VPg-mediated effect on viral translation disturbed the natural stoichiometry of VPg production. Thus, the more PVA RNA is allocated to translation via VPg, the less is available for assembly and movement. Obviously, the correct stoichiometry of viral protein products is a prerequisite for a balanced natural infection. P0 and VPg coexpression increased the steady-state amount of PVA^ΔGDD RNA (Fig. 9A), indicating that replication was not necessary for RNA accumulation. Because the transcriptions of control FLUC and PVA^ΔGDD are both driven by the 35S promoter, the specific effects of P0 and VPg on PVA^ΔGDD RNA translation and accumulation are likely posttranscriptional. The 20- to 100-fold increase in the PVA^ΔGDD expression level is an intriguing example of how extensively posttranscriptional mRNA regulation can affect gene expression. VPg may reroute PVA^ΔGDD RNA from degradation pathways to translation. Although there have been other examples of the interdependence between mRNA translation and degradation (69, 70), it needs to be considered whether our results with PVA^ΔGDD reflect the mechanisms operating during natural PVA infection. This RNA is derived by nuclear transcription and carries a 5′-cap structure instead of 5′-linked VPg present in the authentic viral RNA. The specific effects observed for PVA^ΔGDD compared with our nonviral FLUC control (Fig. 10G and H) suggest that PVA^ΔGDD RNA has virus-specific features that mediate gene expression responses that are at least somewhat similar to those observed with authentic viral RNA (Fig. 9C and 10B) (40). Interestingly, we observed that PVA RNA induces the formation of P0-containing granule structures (A. Hafrén and K. Mäkinen, unpublished results). Similar to wt PVA, PVA^ΔGDD RNA also induced P0- but not P1- or P2-containing granule-like structures in plant cells, whereas the nonviral control FLUC did not. These granules may represent structures associated with viral RNA metabolism, similar to the P bodies and stress granules described, for example, for virus-infected animal cells (71).

Finally, it is noteworthy that silencing of P0 caused abundant accumulation of PVA^CPmut RNA. We are presently unable to distinguish between the contributions of RNA synthesis via replication and RNA degradation to the steady-state PVA RNA level. Moreover, mutant CP may contribute to the different outcomes of P0 silencing for PVA^CPmut and wt PVA. For example, P0 may be an essential host factor required for coordinating PVA RNA functions among translation, replication, degradation, and assembly/movement pathways. In summary, we hypothesize that (i) P0 is an essential host component required to achieve a normal strength of infection and (ii) P0 regulates PVA RNA translation together with VPg and eIF(iso)4E.